codesections There are a lot of ways to do something like that, depending on the exact usecase. From what you've said so far, I'd probably be inclined to zip the two lazy lists together and then diff each line one at a time. See But I'd be happy to give thought to a more specific solution if 00:54
you'd care to share more details (either here or on StackOverflow – everyone who answers/moderates the [raku] tag is very welcoming)
avuserow I thought about suggesting zip with lines, but then you don't have a way to re-synchronize in the case where a line is added. 01:56
<@297037173541175296> did you post the python version somewhere? Curious if I'm understanding the problem correctly 01:57
03:22 zacts left 05:35 codesections left 05:36 codesections joined
Nemokosch I think we have the same thought in mind 07:05
The exact thing I had to do with Python was a bit more convoluted so I'm gonna turn it a bit more theoretical so please treat it as a pseudocode. The lines have a key-value structure conceptually, that's the basis of saying whether one line is extra or they are differing. 07:24
def diff_lines(left_iter, right_iter):
left_line = next(left_iter)
right_line = next(right_iter)
while True:
cmp_result = compare_keys(left_line, right_line)
if cmp_result < 0:
print('Announce extra line on the left side')
left_line = next(left_iter)
elif cmp_result > 0:
print('Announce extra line on the right side')
right_line = next(right_iter)
I think this is enough to grasp the supposed behavior. This solution isn't precise because the left_line can take a step when the right_iter is exhausted so it's possible that we lose one line but that doesn't add to the point 07:26
gfldex `.lines` on a filehandle is a `Seq` so there is your bottled up iterator. `Z` will also return a `Seq` 07:34
m:``` 07:35
for $* Z $* -> [$left, $right] {
say $left, $right;
:-/ 07:36
Nemokosch I know that but how do you tell Z that certain subsequent lines shouldn't be matched against each other but should be considered as extra?
gfldex m: Seq.^methods.grep(*.name eq 'pop').say; 07:38
Nemokosch this is on point
gfldex `Seq` got `.skip` 07:39
Nemokosch without zipping them then, right? 07:40
gfldex Alas, `.skip` is not a mutator. So there is some wrapping required. 07:44
Nemokosch Well this is a bit too cryptic for me 😅 07:46
Actually I think I succeeded with iterators but it was rather painful for exactly one reason: IterationEnd is Mu 07:53
if it was any normal value (literally haha), it would have been no big deal
gfldex ``` 08:02
my @left = $*;
while @left {
@left.shift if (True, False).roll;
try put @left.shift;
<@!297037173541175296> You assign the `Seq` to an `Array` and then use the mutators on the `Array` to manipulate the iterator of the `Seq` indirectly. 08:03
Nemokosch is this lazy? 08:04
gfldex To answer that question would require to understand 08:07
08:07 dakkar joined
It does use `IterationBuffer` so it looks to me as if it would be lazy in chunks. lizmat can likely shed more light on this then me. 08:09
Nemokosch because yes, given the files with like 20k lines, slurping them and doing something afterwards, it would still be okay and dead simple 08:10
lizmat that is lazy
Nemokosch but it's more than just a hypothetical that the files can get big and we have more time than memory in this case 08:11
lizmat also, since $*PROGRAM is an IO::Path, you don't need the .open
Nemokosch interesting that something with an Array interface can be lazy 😮 08:12
gfldex Sadly, I don't have a text file bigger then 64GB to test if it does fill all the RAMz. :) 08:13
Nemokosch 64 GB, heh
I managed to get the Python script that acted upon ~ 100 MB files dead on a 4 GB VM 08:14
probably trying to sort the records real-time wasn't a marvelous idea 🙂 08:15
Morfent m:```
use v6;
.say for Array(gather for 1..3 { .say; .take })
that's not how i test what i wanna test
m:``` 08:16
use v6;
.say for gather for 1..3 { .say; .take }
m:``` 08:17
use v6;
.say for my @ = gather for 1..3 { .say; .take }
m:``` 08:18
use v6;
.say for my @ = lazy gather for 1..3 { .say; .take }
there we go
gfldex We spend great afford to be able to be lazy. :) 08:20
Nemokosch Not gonna lie, it's annoying that all languages I know have iterators some way and they somehow always get to mess it up with a useless interface to it lol 08:21
I'm starting to think the Java interface actually isn't bad
no matter how ugly this hasNext next combo sounds 08:22
with this nice `try` statement prefix, even the Python interface could be decent in Raku 08:23
gfldex The irc-bridge bot basically lives of `try`-statements. :) 08:31
08:31 discord-raku-bot left 08:32 discord-raku-bot joined 11:56 Guest28 joined 11:58 Guest28 left
avuserow So from a higher order perspective, you kind of want a zip function that takes a comparator function to determine whether to consume from the left side, right side, or both. Basically you'd pass in your compare_keys function 13:53
Nemokosch yes, kinda 14:19
16:36 dakkar left